This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: mips gas is horribly broken
On Wed, Jun 06, 2001 at 05:26:42PM -0700, Ian Lance Taylor wrote:
> "H . J . Lu" <hjl@lucon.org> writes:
>
> > On Wed, Jun 06, 2001 at 05:12:43PM -0700, Ian Lance Taylor wrote:
> > >
> > > Here is the comment from bfd_install_relocation:
> > >
> > > entirely the wrong place for such a thing. The second obvious point
> > > is that the current code ignores the reloc addend when producing
> >
> > Am I right to say the main problem with bfd_install_relocation is it
> > doesn't alway do the right thing with the reloc addend?
>
> I would say that the main problem is that it assumes that the object
> file contents are valid, and that the relocation addend is valid for a
> relocateable object file.
What is the best way to provide necessary information to
bfd_install_relocation? I am thinking to add a new field,
addend_info, to reloc_cache_entry. We initialize it to 0
and each backend can use it to pass the necessay information
to bfd_install_relocation.
H.J.